Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 1 | #-*- coding: utf-8 -*- |
| 2 | |
| 3 | import datetime |
Max Rasskazov | 69c2926 | 2015-07-16 19:22:16 +0300 | [diff] [blame^] | 4 | import os |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 5 | |
| 6 | import utils |
| 7 | |
| 8 | from rsync_remote import RsyncRemote |
| 9 | from utils import singleton |
| 10 | |
| 11 | |
| 12 | @singleton |
| 13 | class TimeStamp(object): |
Max Rasskazov | 83e10a5 | 2015-06-18 14:08:12 +0300 | [diff] [blame] | 14 | def __init__(self, now=None): |
| 15 | # now='2015-06-18-104259' |
| 16 | self.snapshot_stamp_format = r'%Y-%m-%d-%H%M%S' |
| 17 | self.snapshot_stamp_regexp = r'[0-9]{4}-[0-9]{2}-[0-9]{2}-[0-9]{6}' |
| 18 | |
| 19 | if now is None: |
| 20 | self.now = datetime.datetime.utcnow() |
| 21 | else: |
| 22 | self.now = datetime.datetime.strptime(now, |
| 23 | self.snapshot_stamp_format) |
| 24 | self.snapshot_stamp = self.now.strftime(self.snapshot_stamp_format) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 25 | |
| 26 | def __str__(self): |
Max Rasskazov | 83e10a5 | 2015-06-18 14:08:12 +0300 | [diff] [blame] | 27 | return self.snapshot_stamp |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 28 | |
Max Rasskazov | d77a5e6 | 2015-06-19 17:47:30 +0300 | [diff] [blame] | 29 | def reinit(self, *args, **kwagrs): |
| 30 | self.__init__(*args, **kwagrs) |
| 31 | |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 32 | |
Max Rasskazov | 3e83758 | 2015-06-17 18:44:15 +0300 | [diff] [blame] | 33 | class TRsync(RsyncRemote): |
Max Rasskazov | 6d7b5c8 | 2015-07-13 13:31:58 +0300 | [diff] [blame] | 34 | # TODO: possible check that rsync url is exists |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 35 | def __init__(self, |
| 36 | rsync_url, |
Max Rasskazov | 0127162 | 2015-06-17 02:35:09 +0300 | [diff] [blame] | 37 | snapshot_dir='snapshots', |
| 38 | latest_successful_postfix='latest', |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 39 | save_latest_days=14, |
| 40 | init_directory_structure=True, |
Max Rasskazov | 83e10a5 | 2015-06-18 14:08:12 +0300 | [diff] [blame] | 41 | timestamp=None, |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 42 | ): |
Max Rasskazov | 3e83758 | 2015-06-17 18:44:15 +0300 | [diff] [blame] | 43 | super(TRsync, self).__init__(rsync_url) |
| 44 | self.logger = utils.logger.getChild('TRsync.' + rsync_url) |
Max Rasskazov | 83e10a5 | 2015-06-18 14:08:12 +0300 | [diff] [blame] | 45 | self.timestamp = TimeStamp(timestamp) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 46 | self.logger.info('Using timestamp {}'.format(self.timestamp)) |
Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 47 | self.snapshot_dir = self.url.a_dir(snapshot_dir) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 48 | self.latest_successful_postfix = latest_successful_postfix |
| 49 | self.save_latest_days = save_latest_days |
| 50 | |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 51 | if init_directory_structure is True: |
| 52 | self.init_directory_structure() |
| 53 | |
| 54 | def init_directory_structure(self): |
Max Rasskazov | 0127162 | 2015-06-17 02:35:09 +0300 | [diff] [blame] | 55 | if self.url.url_type != 'path': |
| 56 | server_root = RsyncRemote(self.url.root) |
| 57 | return server_root.mkdir( |
| 58 | self.url.a_dir(self.url.path, self.snapshot_dir) |
| 59 | ) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 60 | |
Max Rasskazov | 501cea5 | 2015-06-29 18:11:25 +0300 | [diff] [blame] | 61 | def push(self, source, repo_name, symlinks=[], extra=None, save_diff=True): |
Max Rasskazov | 69c2926 | 2015-07-16 19:22:16 +0300 | [diff] [blame^] | 62 | repo_basename = os.path.split(repo_name)[-1] |
Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 63 | latest_path = self.url.a_file( |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 64 | self.snapshot_dir, |
Max Rasskazov | 69c2926 | 2015-07-16 19:22:16 +0300 | [diff] [blame^] | 65 | '{}-{}'.format(self.url.a_file(repo_basename), |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 66 | self.latest_successful_postfix) |
| 67 | ) |
Max Rasskazov | dc8df00 | 2015-06-25 19:37:41 +0300 | [diff] [blame] | 68 | |
| 69 | symlinks = list(symlinks) |
| 70 | symlinks.insert(0, latest_path) |
| 71 | |
Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 72 | snapshot_name = self.url.a_file( |
Max Rasskazov | 69c2926 | 2015-07-16 19:22:16 +0300 | [diff] [blame^] | 73 | '{}-{}'.format(self.url.a_file(repo_basename), self.timestamp) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 74 | ) |
Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 75 | repo_path = self.url.a_file(self.snapshot_dir, snapshot_name) |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 76 | |
Max Rasskazov | 854399e | 2015-06-05 16:35:17 +0300 | [diff] [blame] | 77 | extra = '--link-dest={}'.format( |
Max Rasskazov | 46cc073 | 2015-06-05 19:23:24 +0300 | [diff] [blame] | 78 | self.url.a_file(self.url.path, latest_path) |
Max Rasskazov | 854399e | 2015-06-05 16:35:17 +0300 | [diff] [blame] | 79 | ) |
Max Rasskazov | b1b5083 | 2015-06-18 11:24:53 +0300 | [diff] [blame] | 80 | |
Max Rasskazov | b1b5083 | 2015-06-18 11:24:53 +0300 | [diff] [blame] | 81 | # TODO: split transaction run (push or pull), and |
| 82 | # commit/rollback functions. transaction must has possibility to |
| 83 | # rollback after commit for implementation of working with pool |
| 84 | # of servers. should be something like this: |
| 85 | # transactions = list() |
| 86 | # result = True |
| 87 | # for server in servers: |
| 88 | # transactions.append(server.push(source, repo_name)) |
| 89 | # result = result and transactions[-1].success |
| 90 | # if result is True: |
| 91 | # for transaction in transactions: |
| 92 | # transaction.commit() |
| 93 | # result = result and transactions[-1].success |
| 94 | # if result is False: |
| 95 | # for transaction in transactions: |
| 96 | # transaction.rollback() |
Max Rasskazov | a591185 | 2015-06-17 18:29:58 +0300 | [diff] [blame] | 97 | transaction = list() |
| 98 | try: |
| 99 | # start transaction |
Max Rasskazov | 3e83758 | 2015-06-17 18:44:15 +0300 | [diff] [blame] | 100 | result = super(TRsync, self).push(source, repo_path, extra) |
Max Rasskazov | dc8df00 | 2015-06-25 19:37:41 +0300 | [diff] [blame] | 101 | transaction.append(lambda p=repo_path: self.rmdir(p)) |
Max Rasskazov | a591185 | 2015-06-17 18:29:58 +0300 | [diff] [blame] | 102 | self.logger.info('{}'.format(result)) |
| 103 | |
Max Rasskazov | 501cea5 | 2015-06-29 18:11:25 +0300 | [diff] [blame] | 104 | if save_diff is True: |
| 105 | diff_file = self.tmp.get_file(content='{}'.format(result)) |
| 106 | diff_file_name = '{}.diff.txt'.format(repo_path) |
| 107 | super(TRsync, self).push(diff_file, diff_file_name, extra) |
| 108 | transaction.append(lambda f=diff_file_name: self.rmfile(f)) |
| 109 | self.logger.debug('Diff file {} created.' |
| 110 | ''.format(diff_file_name)) |
| 111 | |
Max Rasskazov | dc8df00 | 2015-06-25 19:37:41 +0300 | [diff] [blame] | 112 | for symlink in symlinks: |
| 113 | try: |
| 114 | tgt = [_[1] for _ in self.ls_symlinks(symlink)][0] |
| 115 | self.logger.info('Previous {} -> {}'.format(symlink, tgt)) |
| 116 | undo = lambda l=symlink, t=tgt: self.symlink(l, t) |
| 117 | except: |
| 118 | undo = lambda l=symlink: self.rmfile(l) |
| 119 | # TODO: implement detection of target relative symlink |
| 120 | if symlink.startswith(self.snapshot_dir): |
| 121 | self.symlink(symlink, snapshot_name) |
| 122 | else: |
| 123 | self.symlink(symlink, repo_path) |
| 124 | transaction.append(undo) |
Max Rasskazov | a591185 | 2015-06-17 18:29:58 +0300 | [diff] [blame] | 125 | |
Max Rasskazov | ad1518a | 2015-06-18 11:24:16 +0300 | [diff] [blame] | 126 | except RuntimeError: |
Max Rasskazov | dc8df00 | 2015-06-25 19:37:41 +0300 | [diff] [blame] | 127 | self.logger.error("Rollback transaction because some of sync" |
| 128 | "operation failed") |
| 129 | [_() for _ in reversed(transaction)] |
| 130 | raise |
| 131 | |
| 132 | try: |
Max Rasskazov | a591185 | 2015-06-17 18:29:58 +0300 | [diff] [blame] | 133 | # deleting of old snapshots ignored when assessing the transaction |
| 134 | # only warning |
Max Rasskazov | dc8df00 | 2015-06-25 19:37:41 +0300 | [diff] [blame] | 135 | self._remove_old_snapshots(repo_name) |
| 136 | except RuntimeError: |
| 137 | self.logger.warn("Old snapshots are not deleted. Ignore. " |
| 138 | "May be next time.") |
Max Rasskazov | a591185 | 2015-06-17 18:29:58 +0300 | [diff] [blame] | 139 | |
Max Rasskazov | 26787df | 2015-06-05 14:47:27 +0300 | [diff] [blame] | 140 | return result |
Max Rasskazov | 452138b | 2015-06-17 02:37:34 +0300 | [diff] [blame] | 141 | |
| 142 | def _remove_old_snapshots(self, repo_name, save_latest_days=None): |
| 143 | if save_latest_days is None: |
| 144 | save_latest_days = self.save_latest_days |
| 145 | if save_latest_days is None or save_latest_days is False: |
| 146 | # delete all snapshots |
| 147 | self.logger.info('Deletion all of the old snapshots ' |
| 148 | '(save_latest_days == {})' |
| 149 | ''.format(save_latest_days)) |
| 150 | save_latest_days = -1 |
| 151 | elif save_latest_days == 0: |
| 152 | # skipping deletion |
| 153 | self.logger.info('Skip deletion of old snapshots ' |
| 154 | '(save_latest_days == {})' |
| 155 | ''.format(save_latest_days)) |
| 156 | return |
| 157 | else: |
| 158 | # delete snapshots older than |
| 159 | self.logger.info('Deletion all of the unlinked snapshots older ' |
| 160 | 'than {0} days (save_latest_days == {0})' |
| 161 | ''.format(save_latest_days)) |
| 162 | warn_date = \ |
| 163 | self.timestamp.now - datetime.timedelta(days=save_latest_days) |
| 164 | warn_date = datetime.datetime.combine(warn_date, datetime.time(0)) |
| 165 | snapshots = self.ls_dirs( |
| 166 | self.url.a_dir(self.snapshot_dir), |
| 167 | pattern=r'^{}-{}$'.format( |
| 168 | repo_name, |
Max Rasskazov | 83e10a5 | 2015-06-18 14:08:12 +0300 | [diff] [blame] | 169 | self.timestamp.snapshot_stamp_regexp |
Max Rasskazov | 452138b | 2015-06-17 02:37:34 +0300 | [diff] [blame] | 170 | ) |
| 171 | ) |
| 172 | links = self.ls_symlinks(self.url.a_dir()) |
| 173 | links += self.ls_symlinks(self.url.a_dir(self.snapshot_dir)) |
| 174 | for s in snapshots: |
| 175 | s_date = datetime.datetime.strptime( |
| 176 | s, |
| 177 | '{}-{}'.format(repo_name, |
Max Rasskazov | 83e10a5 | 2015-06-18 14:08:12 +0300 | [diff] [blame] | 178 | self.timestamp.snapshot_stamp_format) |
Max Rasskazov | 452138b | 2015-06-17 02:37:34 +0300 | [diff] [blame] | 179 | ) |
| 180 | s_date = datetime.datetime.combine(s_date, datetime.time(0)) |
| 181 | s_path = self.url.a_dir(self.snapshot_dir, s) |
| 182 | if s_date < warn_date: |
| 183 | s_links = [_[0] for _ in links |
| 184 | if _[1] == s |
| 185 | or _[1].endswith('/{}'.format(s)) |
| 186 | ] |
| 187 | if not s_links: |
| 188 | self.rmdir(s_path) |
Max Rasskazov | 052e911 | 2015-06-29 18:43:15 +0300 | [diff] [blame] | 189 | self.rmfile(s_path + '.target.txt') |
| 190 | self.rmfile(s_path + '.diff.txt') |
Max Rasskazov | 452138b | 2015-06-17 02:37:34 +0300 | [diff] [blame] | 191 | else: |
| 192 | self.logger.info('Skip deletion of "{}" because there are ' |
| 193 | 'symlinks found: {}'.format(s, s_links)) |
| 194 | else: |
| 195 | self.logger.info('Skip deletion of "{}" because it newer than ' |
| 196 | '{} days'.format(s, save_latest_days)) |